A new TDR probe for measurements of soil solution electrical conductivity

Transcription

1 Manuscript_Moret-Fernández et al JH 1 A new TDR probe for measurements of soil solution electrical conductivity D. Moret-Fernández a *, J. Vicente b, R. Aragüés c, C. Peña a, M.V. López a a Departamento de Suelo y Agua Estación Experimental de Aula Dei, Consejo Superior de Investigaciones Científicas (CSIC). PO Box 0, 000, Zaragoza, Spain b Departamento de Ingeniería Electrónica y Comunicaciones,Universidad de Zaragoza, IA - Edificio I+D, C/ Mariano Esquillor S/N, Despacho.0.0, 001 Zaragoza, Spain. c Centro de Investigación y Tecnología Agroalimentaria de Aragón (CITA)-Diputación General de Aragón (DGA), Unidad de Suelos y Riegos (Unidad Asociada EEAD-CSIC), Avenida Montañana 0, 00 Zaragoza, Spain * Corresponding author. Telf: Fax: address: david@eead.csic.es 1

2 ABSTRACT The measurement of the soil solution electrical conductivity (σ w ) is critical for a better management of irrigation water and the effective monitoring and control of soil salinity. The objective of this work is to present the design and validation of a new time domain reflectometry (TDR) probe (WEC P ) for accurate and non-destructive measurements of σ w. The probe consists in fourteen porous ceramics disks (0. bar bubbling pressure) arranged along the axis of a three-rod TDR probe. Using the Mualem and Friedman () model, σ w was estimated from the volumetric water content (θ) and the bulk electrical conductivity (σ a ) measured in the ceramic disk set of known pore-geometry. The and factors, which describe the complex geometry of the ceramic matrix, were calculated by immersing the probe in NaCl solutions of different electrical conductivities, and in a pressure cell wetted and drained with these NaCl solutions, respectively. The reliability of the WEC P was validated under laboratory and field conditions. The laboratory experiment consisted of the TDR probe inserted in a pressure cell packed with mixed sand and -mm sieved loam soil that was subsequently wetted and drained with different NaCl solutions at various pressure heads. The w estimated by WEC P was compared to the w measured in the draining solutions after they stabilized in the soil porous system. The field experiment compared the w estimated by WEC P with the corresponding w values measured in the soil solution extracted with three ceramic tension lysimeters (TL) after successive wetting and drainage cycles. The and factors calculated for the ceramic disks set were 1. and., respectively. High and significant correlations were found in both laboratory (R = 0.; P < 0.001) and field (R = 0.; P < 0.001) experiments between the σ w estimated by the WEC P and the corresponding σ w values measured in the column-drainage or TL-extracted soil solutions, respectively. These results demonstrate that the WEC P is a feasible instrument to accurately estimate soil solution salinity

3 independently of the soil water content and the porous medium in which the TDR probe is installed. Key words: Water content; Pore-geometry; Bulk electrical conductivity; Time Domain Reflectometry INTRODUCTION Soil salinity, defined as the total concentration of dissolved salts in the soil solution, has a detrimental effect on crops and soil chemical and physical properties (Leone et al., 00). Hence, the accurate measurement of soil salinity is crucial for the productivity and sustainability of irrigated agriculture. Soil salinity is most conveniently measured from the electrical conductivity (EC) of its soil solution (σ w ) (White, 00). Currently, three basic procedures are used to measure or estimate soil salinity (Hendrickx et al., 00): (i) the EC of soil water extracts, (ii) the EC of the soil solution extracted with tension lysimeters, and (iii) the apparent soil bulk EC ( a ) using different methods. The classical soil water extracts using various soil:water ratios, such as the soil saturation extract, are laborious, destructive and impractical when many soil samples are analyzed. The in-situ soil solution extraction method is commonly performed using ceramic tension lysimeters (Parizek and Burke, 10). This low-cost method allows periodic sampling of the soil solution with minimal soil disturbance. Although the tension lysimeters have evolved to new designs (i.e., Wagner, 1; Linden, 1; Hubbell and Sisson, 1), the method is tiresome and limited to soils with relatively high water contents and a proper soil-ceramic contact. Indirect methods to estimate soil salinity are based on the measurement of a determined with electrical resistivity, time domain reflectometry (TDR), or electromagnetic induction techniques. These non-destructive methods

4 estimate w from a and the volumetric soil water content () by using empirical calibration equations or physical based models (Hendrickx et al., 00). The TDR is a non-destructive method that allows real time and simultaneous measurements of the apparent permittivity ( a ), which is related with, and a (Topp and Ferré, 00). The a is calculated from the transit time of the TDR pulse propagating one return trip along a waveguide of length L. Based on the Giese and Tiemann (1) model, σ a is calculated from the attenuation of the long-time reflection coefficient recorded with an uncoated probe (Lin et al., 00). The σ a depends mainly on three variables, effective, σ w, and a geometric factor which accounts for the complex geometry of the soil matrix (Rhoades et al., 1; Mualem and Friedman, ). Several models relating σ a to σ w as a non-linear function of θ have been developed and applied to mineral soils (Rhoades et al., 1; Rhoades et al., 1; Mualem and Friedman, ; Vogeler et al., 1; Persson, 1; Hilhorst, 000; Muñoz-Carpena et al., 00). Persson (00), working with TDR probes installed in sandy soils, showed that the Hilhorst (000) model was as good as other commonly used models for σ w estimates with significant dependency of the linear model on soil type. Mortl et al. (0) compared four equations relating σ w, σ a and θ for three soil series encountered in the floodplain of a southeastern coastal river in USA, and found that the empirical relationship proposed by Vogeler et al. (1) performed the best (overall R = 0. for the three soils), though all models performed satisfactorily in all soils (0. R 0.). Despite all these efforts, σ w can be only consistently predicted from σ a if the relationship between σ w, σ a and θ is known (Hamed et al., 00). Due to variations in responses from different soil types, soil-specific σ w - σ a -θ calibrations are commonly required (Mortl et al., 0). This work presents a new TDR design for accurate and non-destructive estimates of σ w. The TDR probe, which consists in fourteen porous ceramics disks arranged along the axis of a

5 three-rod TDR probe, estimates σ w from θ and σ a measured by TDR in the ceramic disks set. This method is based in the hypothesis that the soil solution is in equilibrium with that in the ceramic disks. Since a constant porous structure is defined inside the ceramic disks, a unique ceramic-specific σ w -σ a -θ calibration is required MATERIAL AND METHODS.1. TDR theory The transit time of the TDR pulse propagating one return trip in a transmission line of length L (m), t L, is expressed by t L ε L a (1) c where c is the speed of light in free space ( x m s -1 ) and a is the apparent permittivity of the medium (Topp and Ferré, 00). The t L value is calculated as the distance between the time at which the signal enters the TDR rods (first peak) and the time when the signal arrives at the end of the TDR probe, also denoted as the second reflection or end point (Heimovaara, 1). Estimations of from a can be calculated by the Topp and Reynolds (1) linear calibration equation: t s θ () tair where t s and t air are the travel time of the TDR pulse propagating along the transmission line when immersed in soil and air, respectively. It is well know that a increases with a (Robinson et al., 00; Evett et al., 00). Assuming that the relaxation effects are negligible, Evett et al. (00) proposed a calibration equation for conventional TDR in terms of a, the travel time, and the effective frequency (f vi, MHz) of the TDR pulse in a probe of length L as:

6 t s σ a θ A B t () air πf viε0 where 0 is the dielectric constant of free space (. -1 F m -1 ) and A and B are empirical factors calculated from a calibration experiment. The reflection coefficient,, as a function of time, t, is typically defined as ρ t V V t 0 V i 0. V () where V(t) is the measured voltage at time t, V 0 is the voltage in the cable just prior to the insertion of the probe (standard impedance value of 0 ), and V i is the incident voltage of the cable tester prior to the pulse rise. The soil bulk electrical conductivity ( a ) estimated with the long-time analysis of the TDR waveform is calculated using the Giese and Tiemann (1) equation: K p 1 ρ,scale σ a () Zr 1 ρ,scale where Z r is the output impedance of the TDR cable tester (0 Ω) and K p (m -1 ) is the probe- geometry-dependent cell constant value which can be calculated from the characteristics of the TDR probe geometry (Evett et al., 00), or by immersing the probe in different electrolyte solutions of known EC (Wraith, 00). The ρ, Scale is the scaled steady-state reflection coefficient corresponding to the ideal condition in which there is no instrument error or cable resistance. The ρ, Scale is calculated using the equation described by Lin et al. (00): 1 1 ( air SC )( air ) )( ) ( )(1, Scale (1 SC air air SC air ) 1 () 0 where, air and SC are the long-time reflection coefficients measured in the studied 1 medium, in air and in a short-circuited probe, respectively.

7 Soil solution electrical conductivity ( w ) estimation Following the hypothesis proposed by Mualem and Friedman (), which assume that the tortuosity factor affecting the soil bulk electrical conductivity ( a ) is identical to that defined for predicting the soil hydraulic conductivity, a () can be expressed as: σ a θ β θ σa σ sat as θ () sat where a-sat and sat are the soil bulk electrical conductivity and the volumetric soil water content at saturation, respectively, a-s is the bulk electrical conductivity of the soil solid phase, and is as factor that depends on the soil water transmission porosity and defines the decrease rate between a and. According to Mualem and Friedman (), a-sat can be defined as: σ σ θ τ a sat w sat () where is a transmission coefficient at soil water saturation that describes the tortuous nature of the current lines that decreases the mobility of ions near the soil-liquid and liquid-gas interfaces. Taking the hypothesis that w only depends on the dissolved salts (Rhoades et al., 1), w could be theoretically estimated by combining equations () and () as w sat a sat as () 0 1 w, which depends on temperature, T ºC, was corrected to ºC ( w/ ) according to (Rhoades et al., 1) w/ = w * f ()

8 where f is an empirical factor expressed as (US Salinity Laboratory Staff, 1 ) and T = (T ºC -)/ f = (T) (T ) (T ), () TDR probe designs All TDR measurements were performed using a TDR0 (Campbell Sci.) model cable tester. A 1.0-m 0- coaxial cable directly connected the TDR probes to the TDR pulser. The TDR waveforms were transferred to a computer for display and analysis using the software TDR-Lab V.1.0. (Moret-Fernández et al., 0), which automatically calculates a and a. The TDR probe used to estimate the soil water pore electrical conductivity (WEC P ) is similar to the design developed by Or and Wraith (1) for measuring the soil matric potential. This consists in fourteen disks (-mm thick and 0-mm in diameter) of commercially available porous ceramics plates with a bubbling pressure of -0. bar (Soil Moisture Inc. UK). The disks were arranged along the axis of a three-rod TDR probe (rod length: 1. mm; rod diameter:. mm; spacing of the outer conductors: 0.0 mm). A second three-rod TDR probe without the ceramic disks (rod length: 0. mm; rod diameter:. mm; spacing of the outer conductors: 0. mm) for soil and a estimations was also made (SWC P ). In both cases, a cm length coaxial cable connected the three-rods of the TDR probe to a male-bnc connector. The head of the two TDR probes (-cm height) was made of a commercial available epoxy resin. 1.. Laboratory calibration and validation experiments A laboratory experiment was performed to calculate the K p values of WEC P and SWC P. This was experimentally estimated from Eq. () by immersing the WEC P (without ceramic disks) and SWC P in cylindrical plastic containers (00 mm internal diameter -i.d.-, and 00

9 mm height) filled with six NaCl solutions of EC = 0., 1,,, and 1 ds m -1. The EC was measured with a Crison conductimeter model, and all values were corrected to ºC (Eq. ). A new series of laboratory experiments were performed to calculate the and coefficients (Mualem and Friedman, ) (Eqs. and ) of the WEC P ceramic disks. The coefficient (Eq. ) was calculated in a column experiment, in which the WEC P inserted in the ceramic disks was located in the plastic containers. A first measurement of and a was done with the ceramic disks dry. Next, the WEC P was immersed in the container filled with distilled water and and a were recorded h later. This procedure was repeated using the previous six NaCl solutions. In all cases, a previously free salts WEC P was used. The coefficient was numerically calculated by minimizing the Root Mean Square Error (RMSE) between the TDR-measured a (Eq. ) and the calculated a-sat (Eq. ) for an average sat. The factor was estimated in a subsequent laboratory experiment in which the WEC P was located in a pressure cell. This consisted of a plastic tube (1. mm i.d. and.0 mm height) closed at both ends with two plastic lids (1. mm i.d. and.0 and. mm height for the top and bottom lids, respectively) drilled with a single hole. Two rubber joints placed between the lids and the plastic tube hermetically closed the pressure cell. A first measurement of and a was done with the ceramic disks dry. Next, the WEC P was saturated by injecting a ds m -1 NaCl solution through the base of the pressure-cell. The ceramic disks were considered saturated and equilibrated with the NaCl solution when it exited the top of the pressure cell with the same EC then that used to saturate the WEC P. This process took approximately hours. Next, the ceramic disks of the WEC P were sequentially desaturated at different pressure heads (,,, 0 and 0 kpa) by injecting air through the top of the pressure cell. The extracted water was collected and its EC was measured. The values of and a were recorded

10 at soil saturation and hours following each pressure-head step. This experiment was repeated twice using a ds m -1 NaCl solution. Finally, assuming a negligible a-s (Eq. ), the factor was numerically calculated by minimizing the RMSE between the measured w and the estimated w (Eq. ), for an average sat. This TDR probe was validated in a pressure cell laboratory experiment and under field conditions. The pressure cell consisted of a plastic tube (0 mm i.d., 0 mm height) with a mm i.d. hole drilled at mm height, and closed at the ends with two plastic lids (Fig. 1). The bottom lip had inserted a 0. bar ceramic plate (-mm thick and 0-mm in diameter) (Soil Moisture Inc. UK), which was placed on a mm i.d. hole. These two holes allowed the flow of air and water during the soil wetting and draining processes. Two female-female BNC connectors, in which the WEC P and SWC P were connected, were inserted though the top lip. A thermocouple was also inserted in the pressure cell for soil temperature measurements. The cell was filled up and uniformly packed with sand (0 m grain size) until the head of the TDR probes were half covered. Next, a ds m -1 NaCl solution was slowly injected through the base of the pressure cell until the EC of the outlet solution equalled the inlet one ( hours approximately). The total volume of water added was approximately four times the total soil porosity. Once the sand was saturated and equilibrated with the NaCl solution, the column was sequentially drained at pressure heads of 0.,,,, 0 and 0 kpa, by injecting air through the lateral pressure cell hole. The water drained at each pressure head was collected and the corresponding EC measured. Values of and a obtained with the WEC P and SWC P were recorded at soil saturation and hours after imposing each pressure head. According to Eq. (), w was calculated from the measured and a values and the and factors estimated in the previous experiments. The w values were corrected to ºC (Eq. ). This experiment was repeated using a -mm sieved loam soil saturated with three different KCl solutions of,

11 and ds m -1. Finally, the TDR-estimated w values were statistically compared to the measured EC values in the inlet solutions Field testing The field experiment consisted in comparing the w estimated by WEC P to the EC values measured in the soil solution extracted with ceramic tension lysimeters (TL). The experiment was performed on a loam soil located in an apple orchard of the Estación Experimental de Aula Dei (Zaragoza). The soil bulk density was 1. g cm -. Three TL (model SPS 00 - SDEC) were inserted into the soil at the vertices of a 1 cm equilateral triangle, the WEC P was inserted in the center of the triangle, and the SWC P at a cm distance from the WEC P. Both TDR probes were inserted at the same depth that the TL. The heads of the two TDR probes were buried 1 cm under the soil surface. The experimental plot was confined in a 0 cm diameter and 0 cm height plastic tube driven 1 cm into the soil. Successive soil wettingdrainage cycles were repeated with distilled water and KCl-water solutions of different EC (Table 1) until soil equilibrium. Systematic measurements of and a were recorded with the WEC P and SWC P, and the soil solution was extracted with the TL for the measurement of w. Soil temperatures were measured with a thermocouple sensor installed at cm depth inside the experimental plot. The average w measured in the solutions extracted with the three TLs were compared to the corresponding w values estimated with the WEC P from the recorded and a values (Eq. ). All w were corrected at ºC 1. RESULTS AND DISCUSION The K p value of WEC P without the ceramic disks and SWC P estimated from the laboratory experiment were. and. m -1, respectively. The A and B empirical factors of Eq. ()

12 applied to the WEC P and SWC P to calculate the volumetric water content corresponded to the respective 0.1 and 0. values given by Topp and Reynolds (1) (Eq. ). The average sat used in Eq. () and () to calculate the and factors and estimate w was 0. cm cm -. The factor (Eq. ) obtained from the laboratory experiment under saturated conditions was 1.. This value, which was slightly higher than the 1. value proposed by Mualem and Friedman () for coarse-textured soils, allowed an excellent correlation (p < 0.001) between the a-sat measured by WEC P and calculated with Eq. (Fig. ). The value (Eq. ), calculated from the pressure cell experiments was.. This value, almost twice higher than the. value reported by Mualem and Friedman () for coarse-textured soils, also allowed an excellent (p < 0.001) correlation between the a measured by WEC P and calculated with Eq. (Fig. ). Exponential relationships were found between and a measured with the WEC P (Fig. ) in the pressure cell and the sand and -mm sieved loam soil columns experiments. As described by the Mualem and Friedman () model, a exponentially decreases with decreases in, and the - a slopes get smoother with decreasing w values (Fig. ). Finally, a noble correlation (p < 0.001) was found between the w measured in all the column experiments (water, pressure cell, sand and loam soil) and the corresponding w estimates with the WEC P (Eq. ) for estimated, a, sat and and factors (Eq. and ) (Fig. ). Figure shows the time-evolution of and a measured with SWC P and WEC P and w estimated with WEC P (Eq. ) in the sand and -mm sieved loam soil column after being saturated with solutions of, and ds m -1 EC, and subsequently drained at pressure heads ranging between and 0 kpa. The and a values measured with the two TDR probes decreased with increasing pressure heads, but the decrease was in general much smaller with WEC P than with SWC P. As shown in Fig., the amplitude of a as a function of increases 1

13 with increasing solution EC. Important differences in a measured with the SWC P were observed between the sand and the loam soil columns. This should be attributed to the different and factors of these porous media. Thus, the and factors approached from the and a measured in the laboratory experiments with the SWC P were 1. and 1. for the sand and.0 and 1. for the -mm sieved loam soil, respectively. The most relevant result shown in Fig. is that w estimated with the WEC P using Eq. was independent of and the porous media in which the probe was inserted, and that it was similar to the w imposed with the different NaCl or KCl solutions. These results indicate that the new TDR probe is a feasible method for accurate and non-destructive estimates of soil solution EC for the porous media and pressure heads examined in this work. The results obtained in the laboratory experiments were supported by those obtained under field conditions where the w values estimated with the WEC P were compared to those measured in the soil solutions extracted with the three tension lysimeters (TL). Overall, an excellent correlation (P < 0.001) was observed between the TDR-estimated w and the TLmeasured w, with a regression coefficient not significantly different from one (Fig. ). The dynamics of and a measured with SWC P and WEC P and the w measured with the TL and estimated with the WEC P were similar to those observed in the laboratory. While a was in all cases dependent on and on the EC of the infiltrating solution, w estimated with the WEC P was only dependent on the EC of the infiltrating solution. Hence, w did not change appreciably with time (i.e., with decreases in ), in contrast with the observed sharp decreases of a with time (Fig. ). An increase of w was observed when the KCl solutions were added in subsequent events to the soil, so that they were similar to the w measured in the soil solution extracted by the TL. Similarly, the WEC P estimated w and the TL-measured w were also similar during the leaching process (i.e., addition of distilled water in cumulative days 1

14 and ), except in the hrs following the application of distilled water (Fig. ). These results suggest that the WEC P needs almost two days to equilibrate the solution within the ceramic discs with the solution within the soil pores. This response time of the WEC P is not a relevant handicap for the long-term assessment of soil salinity CONCLUSIONS This work presents a new TDR probe (WEC P ) to estimate the soil solution electrical conductivity. The design, consisting in a three-rod TDR probe embedded in fourteen porous ceramics disks, is based in the hypothesis that the solution in the ceramic disks equilibrates with the soil solution present in the soil pores. Since the ceramic disks have a constant porousgeometry, a unique ceramic-specific σ w -σ a -θ calibration is required. The new probe was calibrated and subsequently validated in laboratory and field experiments. The results demonstrate that the new TDR probe allows accurate estimates of soil solution EC (σ w ) independently of the soil water contents imposed in these experiments. Although the TDR equipment used in these experiments is relatively expensive, the large versatility of this technique, which allows working with homemade TDR probes, allows achieving a return on the investment. Some advantages of this new design of TDR probe can be summarized as: (a) low cost sensor (made from a simple TDR probe and commercial available ceramic discs); (b) quick and easy field installation; and (c) robustness and low maintenance cost. However, further efforts should be done to (i) incorporate a temperature sensor that will correct σ w to a reference temperature of ºC, (ii) use alternative porous media to estimate σ w at higher pressure heads while minimizing the response time to changes in the external soil solution, (iii) improve the TDR probe design to allow simultaneous estimates of σ w and the soil matric potential, and (iv) include the Mualem and Friedman () model, or similars to estimate σ w 1

15 in available TDR software (i.e. TDR-Lab) for faster estimates of the soil solution electrical conductivity. Acknowledgements This research was supported by the Comisión Interministerial de Ciencia y Tecnología of Spain (Grant AGL0-00-CO-0/AGR). C. Peña was awarded with a FPI fellowship by the Spanish Ministry of Science and Innovation. The authors are grateful to Ricardo Gracia, M. Josefa Salvador and Ana Bielsa for their help in various technical aspects of this study REFERENCES Evett, S.R., Tolk, J.A., Howell, T.A. 00. Soil profile water content determination: Sensor accuracy, axial response, calibration, temperature dependence, and precision. Vadose Zone Journal, -0. Evett, SR, Tolk, J.A., Howell, T.A. 00. Time domain reflectometry laboratory calibration in travel time, bulk electrical conductivity, and effective frequency. Vadose Zone Journal, 0-. Giese, K., Tiemann, R. 1. Determination of the complex permittivity from thin-sample time domain reflectometry: Improved analysis of the step response waveform. Advances in Molecular Relaxation Processes, -. Hamed, Y., Samy, G., M., Persson. 00. Evaluation of the WET sensor compared to time domain reflectometry. Hydrological Sciences Journal 1, 1-1. Heimovaara, T.J. 1. Design of triple-wire Time-Domain Reflectometry probes in practice and theory. Soil Science Society of America Journal, 1-. 1

16 Hendrickx J.M.H., Wraith J.M., Crown, D.L., Kachanoski, R. 00. Miscible solute transport. In, Methods of Soil Analysis. Part. (Ed. J.H. Dane and G.C. Topp), SSSA Book Series No.. Soil Science Society of America, Madison WI. Hilhorst, M A pore water conductivity sensor. Soil Science Society of America Journal, 1-1 Hubbell, J.M., Sisson J.B. 1. Portable tensiometer use in deep boreholes. Soil Science, 1. Leone, AP., Menenti, M, Buondonno, A., Letizia, A., Maffei, C, Sorrentino, G. 00. A field experiment on spectrometry of crop response to soil salinity. Agricultural Water Management, -. Lin, C-P., Chung, C.C., Huisman, J.J., Tang, S.H. 00. Clarification and calibration of reflection coefficient for electrical conductivity measurement by Time Domain Reflectometry. Vadose Zone Journal : -0. Linden, D.R. 1. Design, installation and use of porous ceramic samplers for monitoring soil water quality. USDA Agric. Res. Serv. Tech. Bull. 1. Moret-Fernández, D., Vicente, J., Lera, F., Latorre, B., López, M.V., Blanco, N., González- Cebollada, C., Arrúe, J.L., Gracia, R., Salvador, M.J., Bielsa, A., 0. TDR-Lab Version 1.0 Users Guide ( Mortl, A., Muñoz-Carpena, R., Kaplan, D., Li, Y. 0. Calibration of a combined dielectric probe for soil moisture and porewater salinity measurement in organic and mineral coastal wetland soils. Geoderma, 0- Mualem, Y., Friedman, S.P.. Theoretical prediction of electrical conductivity in saturated and unsaturated soil. Water Resources Research : 1-. 1

17 Muñoz-Carpena, R., Regalado, C.M., Ritter, A., Álvarez-Benedí, J. Socorro, A.R. 00. TDR estimation of electrical conductivity and saline solute concentration in a volcanic soil. Geoderma, 1, - 1. Or, D., Wraith, J.M. 1. A new soil matric potential sensor based on time domain reflectometry. Water Resources Research, -0. Parizek, R.R. and Burke E.L. 10. Soil-water sampling using pan and deep pressure-vacuum lysimeters. Journal of Hydrology, 1-1. Persson M. 1. Soil solution electrical conductivity measurements under transient conditions using time domain reflectometry. Soil Science Society of America Journal. 1, -0. Persson, M. 00. Evaluating the linear dielectric constant-electrical conductivity model using time-domain reflectometry. Hydrological Sciences Journal, -. Rhoades, J.D., Chanduvi, F., Lesch, S. 1. Soil salinity assessment. Methods and interpretation of electrical conductivity measurements. FAO Irrigation and drainage paper,, ISSN 0-. Rhoades, J.D., Manteghi, N.A., Shouse, P.J., Alves, W.J. 1. Soil electrical conductivity and soil salinity: new formulations and calibrations. Soil Science Society of America Journal, -. Rhoades, J.D., Manteghi, N.A., Shouse, P.J., Alves, W.J., 1 Soil electrical conductivity and soil salinity:new formulations and calibration. Soil Science Society of America Journal : -. Robinson, D.A., Jones, S.B., Wraith, J.M. 00. A Review of Advances in Dielectric and Electrical Conductivity Measurement in Soils Using Time Domain Reflectometry. Vadose Zone Journal, -. 1

18 Topp, G.C., and T.P.A. Ferre, 00. Water content, In, Methods of Soil Analysis. Part. (Ed. J.H. Dane and G.C. Topp), SSSA Book Series No.. Soil Science Society of America, Madison WI. Topp, G.C., and W.D. Reynolds. 1. Time domain reflectometry: A seminal technique for measuring mass and energy in soil. Soil Tillage Research, 1-1. US Salinity Laboratory Staff. 1. Diagnosis and improvement of saline and alkali soils. USDA Handbook 0, U.S. Government Printing Office, Washington, D. C. Vogeler, I., Clothier, B.E., Green, S.R., Scotter, D.R., Tillman, R.W. 1. Characterizing water and solute movement by TDR and disk permeametry. Soil Science Society of America Journal 0, -1. Wagner, G.H. 1. Use of porous ceramic cups to sample soil water within the profile. Soil Science,, -. White, RE. 00. Soils for fine wines. Oxford University Press. Wraith, J.M. 00. Time Domain Reflectometry. p In J.H. Dane and G.C. Topp (eds.) Methods of Soil Analysis: Part IV Physical Methods. SSSAJ Madison, WI

19 Figure captions Figure 1. Diagram of the pressure cell used to validate the TDR probe to estimate the water solution electrical conductivity (WEC P ). SWC P denotes a -cm long standard TDR probe. Figure. Relationship and linear regression equation between a-sat measured by TDR and model-calculated a-sat with Eq. using the optimized factor obtained with the WEC P from the column experiments under saturated conditions. 1 1 Figure. Relationship and linear regression equation between a measured by TDR and model-calculated a with Eq. using the optimized and factors and the averaged sat from all the column experiments Figure. Relationships between a and model-calculated with Eq. (lines) and measured with the WEC P (circles) obtained from the pressure cell, sand and loam soil column experiments using three NaCl solutions of, and ds m -1 ECs Figure. Relationship and linear regression equation between w CC measured in all the column experiments (water, pressure cell, sand and loam soil) and w estimated with WEC P ( w TDR) using Eq. (). Figure. Time evolution of a and measured with SWC P and WEC P, and w estimated with WEC P in the sand and loam soil column experiments after being saturated with solutions 1

20 of, and ds m -1 EC (right Y-axis), and subsequently drained at pressure heads ranging between and 0 kpa. Figure. Relationship and linear regression equation between the average soil solution EC measured in the solutions extracted with the three tension lysimeters ( w TL) and the corresponding w values estimated with the WEC P ( w TDR). The horizontal segments denote ± one standard deviation of the mean w TL. 1 1 Figure. Time evolution of soil temperature, and a measured with SWC P and WEC P, w estimated with WEC P, and mean w measured in the soil solutions extracted with the three tension lysimeters (TL). The cumulative days at which the solutions of a given EC were added to the soil are also shown in the bottom figure. The vertical segments denote ± one standard deviation of the mean w TL

21 Manuscript_Moret-Fernández et al JH_Marked Click here to view linked References 1 A new TDR probe for measurements of soil solution electrical conductivity D. Moret-Fernández a *, J. Vicente b, R. Aragüés c, C. Peña a, M.V. López a a Departamento de Suelo y Agua Estación Experimental de Aula Dei, Consejo Superior de Investigaciones Científicas (CSIC). PO Box 0, 000, Zaragoza, Spain b Departamento de Ingeniería Electrónica y Comunicaciones,Universidad de Zaragoza, IA - Edificio I+D, C/ Mariano Esquillor S/N, Despacho.0.0, 001 Zaragoza, Spain. c Centro de Investigación y Tecnología Agroalimentaria de Aragón (CITA)-Diputación General de Aragón (DGA), Unidad de Suelos y Riegos (Unidad Asociada EEAD-CSIC), Avenida Montañana 0, 00 Zaragoza, Spain * Corresponding author. Telf: Fax: address: david@eead.csic.es 1

22 ABSTRACT The measurement of the soil solution electrical conductivity (σ w ) is critical for a better management of irrigation water and the effective monitoring and control of soil salinity. The objective of this work is to present the design and validation of a new time domain reflectometry (TDR) probe (WEC P ) for accurate and non-destructive measurements of σ w. The probe consists in fourteen porous ceramics disks (0. bar bubbling pressure) arranged along the axis of a three-rod TDR probe. Using the Mualem and Friedman () model, σ w was estimated from the volumetric water content (θ) and the bulk electrical conductivity (σ a ) measured in the ceramic disk set of known pore-geometry. The tortuosity and factors, which (), describinge the complex geometry of the ceramic matrix,x, was were calculated by immersing the probe in NaCl solutions of different electrical conductivities, and. The factor, which depends on the soil water transmission porosity, was estimated in a pressure cell wetted and drained with these NaCl solutions, respectively. The reliability of the WEC P was validated under laboratory and field conditions. The laboratory experiment consisted of the TDR probe inserted in a pressure cell packed with mixed sand and -mm sieved loam soil that was subsequently wetted and drained with different NaCl solutions at various pressure heads. The w estimated by WEC P was compared to the w measured in the draining solutions after they stabilized in the soil porous system. The field experiment compared the w estimated by WEC P with the corresponding w values measured in the soil solution extracted with three ceramic tension lysimeters (TL) after successive wetting and drainage cycles. The and factors calculated for the ceramic disks set were 1. and., respectively. High and significant correlations (R = 0.; P < 0.001) were found in both laboratory (R = 0.; P < 0.001) and field (R = 0.; P < 0.001) experiments between the σ w estimated by the WEC P and the corresponding σ w values measured in the column-drainage or TL-extracted soil

23 solutions, respectively. These results demonstrate that the WEC P is a feasible instrument to accurately estimate soil solution salinity independently of the soil water content and the porous medium in which the TDR probe is installed. Key words: Water content; Pore-geometry; Bulk electrical conductivity; Time Domain Reflectometry INTRODUCTION Soil salinity, defined as the total concentration of dissolved salts in the soil solution, has a detrimental effect on crops and soil chemical and physical properties (Leone et al., 00). Hence, the accurate measurement of soil salinity is crucial for the productivity and sustainability of irrigated agriculture. Soil salinity is most conveniently measured from the electrical conductivity (EC) of its soil solution (σ w ) (White, 00). Currently, three basic procedures are used to measure or estimate soil salinity (Hendrickx et al., 00): (i) the EC of soil water extracts, (ii) the EC of the soil solution extracted with tension lysimeters, and (iii) the apparent soil bulk EC ( a ) using different methods. The classical soil water extracts using various soil:water ratios, such as the soil saturation extract, are laborious, destructive and impractical when many soil samples are analyzed. The in-situ soil solution extraction method is commonly performed using ceramic tension lysimeters (Parizek and Burke, 10). This low-cost method allows periodic sampling of the soil solution with minimal soil disturbance. Although the tension lysimeters have evolved to new designs (i.e., Wagner, 1; Linden, 1; Hubbell and Sisson, 1), the method is tiresome and limited to soils with relatively high water contents and a proper soil-ceramic contact. Indirect methods to estimate soil salinity are based on the measurement of a determined with electrical resistivity, time domain reflectometry (TDR), or electromagnetic induction techniques. These non-destructive methods

24 estimate w from a and the volumetric soil water content () by using empirical calibration equations or physical based models (Hendrickx et al., 00). The TDR is a non-destructive method that allows real time and simultaneous measurements of the apparent permittivity ( a ), which is related with, and a (Topp and Ferré, 00). The a is calculated from the transit time of the TDR pulse propagating one return trip along a waveguide of length L. Based on the Giese and Tiemann (1) model, σ a is calculated from the attenuation of the long-time reflection coefficient recorded with an uncoated probe (Lin et al., 00). The σ a depends mainly on three variables, effective, σ w, and a geometric factor which accounts for the complex geometry of the soil matrix (Rhoades et al., 1; Mualem and Friedman, ). Several models relating σ a to σ w as a non-linear function of θ have been developed and applied to mineral soils (Rhoades et al., 1; Rhoades et al., 1; Mualem and Friedman, ; Vogeler et al., 1; Persson, 1; Hilhorst, 000; Muñoz-Carpena et al., 00). Persson (00), working with TDR probes installed in sandy soils, showed that the Hilhorst (000) model was as good as other commonly used models for σ w estimates with significant dependency of the linear model on soil type. Mortl et al. (0) compared four equations relating σ w, σ a and θ for three soil series encountered in the floodplain of a southeastern coastal river in USA, and found that the empirical relationship proposed by Vogeler et al. (1) performed the best (overall R = 0. for the three soils), though all models performed satisfactorily in all soils (0. R 0.). Despite all these efforts, σ w can be only consistently predicted from σ a if the relationship between σ w, σ a and θ is known (Yasser Hamed et al., 00). Due to variations in responses from different soil types, soilspecific σ w -σ a -θ calibrations are commonly required (Mortl et al., 0). This work presents a new TDR design for accurate and non-destructive estimates of σ w. The TDR probe, which consists in fourteen porous ceramics disks arranged along the axis of a

25 three-rod TDR probe, estimates σ w from θ and σ a measured by TDR in the ceramic disks set. This method is based in the hypothesis that the soil solution is in equilibrium with that in the ceramic disks. Since a constant porous structure is defined inside the ceramic disks, a unique ceramic-specific σ w -σ a -θ calibration is required MATERIAL AND METHODS.1. TDR theory The transit time of the TDR pulse propagating one return trip in a transmission line of length L (m), t L, is expressed by t L L εa (1) c where c is the speed of light in free space ( x m s -1 ) and a is the apparent permittivity of the medium (Topp and Ferré, 00). The t L value is calculated as the distance between the time at which the signal enters the TDR rods (first peak) and the time when the signal arrives at the end of the TDR probe, also denoted as the second reflection or end point (Heimovaara, 1). Estimations of from a can be calculated by the Topp and Reynolds (1) linear calibration equation: t s θ () tair where t s and t air are the travel time of the TDR pulse propagating along the transmission line when immersed in soil and air, respectively. It is well know that a increases with a (Robinson et al., 00; Evett et al., 00). Assuming that the relaxation effects are negligible, Evett et al. (00) proposed a calibration equation for conventional TDR in terms of a, the travel time, and the effective frequency (f vi, MHz) of the TDR pulse in a probe of length L as:

26 t s σ a θ A B t () air πf viε0 where 0 is the dielectric constant of free space (. -1 F m -1 ) and A and B are empirical factors calculated from a calibration experiment. The reflection coefficient,, as a function of time, t, is typically defined as ρ t V V t 0 V i 0. V () where V(t) is the measured voltage at time t, V 0 is the voltage in the cable just prior to the insertion of the probe (standard impedance value of 0 ), and V i is the incident voltage of the cable tester prior to the pulse rise. The soil bulk electrical conductivity ( a ) estimated with the long-time analysis of the TDR waveform is calculated using the Giese and Tiemann (1) equation: K p 1 ρ,scale σ a () Zr 1 ρ,scale where Z r is the output impedance of the TDR cable tester (0 Ω) and K p (m -1 ) is the probe- geometry-dependent cell constant value which can be calculated from the characteristics of the TDR probe geometry (Evett et al., 00), or by immersing the probe in different electrolyte solutions of known EC (Wraith, 00). The ρ, Scale is the scaled steady-state reflection coefficient corresponding to the ideal condition in which there is no instrument error or cable resistance. The ρ is calculated using the equation described by Lin et al. (00):, Scale ( air SC )( air ) )( ) ( )(1, Scale (1 SC air air SC air ) ( air SCair )(), Scale 1 1()()( 1)( ) () SCair air SCair 1

27 where, air and SC are the long-time reflection coefficients measured in the studied medium, in air and in a short-circuited probe, respectively Soil solution electrical conductivity ( w ) estimation Following the hypothesis proposed by Mualem and Friedman (), which assume that the tortuosity factor affecting the soil bulk electrical conductivity ( a ) is identical to that defined for predicting the soil hydraulic conductivity, a () can be expressed as: σ a θ β θ σa σ sat as θ () sat where a-sat and sat are the soil bulk electrical conductivity and the volumetric soil water content at saturation, respectively, a-s is the bulk electrical conductivity of the soil solid phase, and is as factor that depends on the soil water transmission porosity and defines the decrease rate between a and. According to Mualem and Friedman (), a-sat can be defined as: σ σ a sat θ τ w sat () where is a transmission coefficient at soil water saturation that describes the tortuous nature of the current lines that decreases the mobility of ions near the soil-liquid and liquid-gas interfaces. Taking the hypothesis that w only depends on the dissolved salts (Rhoades et al., 1), it w could be theoretically estimated by combining equations () and ()from: as w sat a sat as Formatted: Line spacing: Double ()

28 w, which depends on temperature, T (ºC), was corrected to ºC ( w/ ) according to as given by the empirical equation: (Rhoades et al., 1) w/ = w * f () where f is an empirical factor expressed as (US Salinity Laboratory Staff, 1 ) f = (T) (T ) (T ), () and T = (T ºC -)/ w = T T R = 0. ().. Design, TDR probe designs calibration and validation of the TDR probe All TDR measurements were performed using a TDR0 (Campbell Sci.) model cable tester. A 1.0-m 0- coaxial cable directly connected the TDR probes to the TDR pulser. The TDR waveforms were transferred to a computer for display and analysis using the software TDR-Lab V.1.0. (Moret-Fernández et al., 0), which automatically calculates a and a. The TDR probe used to estimate the soil water pore electrical conductivity (WEC P ) is similar to the design developed by Or and Wraith (1) for measuring the soil matric potential. This consists in fourteen disks (-mm thick and 0-mm in diameter) of commercially available porous ceramics plates with a bubbling pressure of -0. bar (Soil Moisture Inc. UK). The disks were arranged along the axis of a three-rod TDR probe (rod length: 1. mm; rod diameter:. mm; spacing of the outer conductors: 0.0 mm). A second three-rod TDR probe without the ceramic disks (rod length: 0. mm; rod diameter:. mm; spacing of the outer conductors: 0. mm) for soil and a estimations was also made Formatted: Font: Italic Formatted: Font: Italic, Subscript Formatted: Subscript Formatted: Indent: Left:. cm, Line spacing: Double Formatted: English (U.K.) Formatted: Font: Italic Formatted: Font: Not Italic Formatted: Indent: First line: 0 cm, Line spacing: Double Formatted: Font: Italic Formatted: Font: Not Italic Formatted: Font: 1 pt, French (France) Formatted: Indent: Left:. cm Formatted: Font: 1 pt, Italic, French (France) Formatted: Font: 1 pt, Italic, French (France), Superscript Formatted: Font: 1 pt, Italic, French (France) Formatted: Font: 1 pt, Italic, French (France), Superscript Formatted: Font: 1 pt, Italic, French (France) Formatted: Font: Not Italic Formatted: Font: Not Italic Formatted: French (France) Formatted: English (U.K.) Formatted: Indent: First line: 0 cm, Line spacing: Double Formatted: Subscript Formatted: Font: Not Italic, English (U.K.), Not Superscript/ Subscript Formatted: Font: Not Italic Formatted: English (U.K.) Formatted: Font: Not Italic Formatted: Line spacing: Double Formatted: Line spacing: Double Formatted: French (France)

29 (SWC P ). In both cases, a cm length coaxial cable connected the three-rods of the TDR probe to a male-bnc connector. The head of the two TDR probes (-cm height) was made of a commercial available epoxy resin Laboratory calibration and validation experiments A laboratory experiment was performed to calculate the K p values of WEC P and SWC P. This was experimentally estimated from Eq. () by immersing the WEC P (without ceramic disks) and SWC P in cylindrical plastic containers (00 mm internal diameter -i.d.-, and 00 mm height) filled with six NaCl solutions of EC = 0., 1,,, and 1 ds m -1. The EC was measured with a Crison conductimeter model, and all values were corrected to ºC (Eq. ). A new series of laboratory experiments were performed to calculate the and coefficients (Mualem and Friedman, ) (Eqs. and ) of the WEC P ceramic disks. The coefficient (Eq. ) was calculated in a column experiment, in which the WEC P inserted in the ceramic disks was located in the plastic containers. A first measurement of and a was done with the ceramic disks dry. Next, the WEC P was immersed in the container filled with distilled water and and a were recorded h later. This procedure was repeated using the previous six NaCl solutions. In all cases, a previously free salts WEC P was used. The coefficient was numerically calculated by minimizing the Root Mean Square Error (RMSE) between the TDR-measured a (Eq. ) and the calculated a-sat (Eq. ) for an average sat. The factor was estimated in a subsequent laboratory experiment in which the WEC P was located in a pressure cell. This consisted in of a plastic tube (1. mm i.d. and.0 mm height) closed at both ends with two plastic lids (1. mm i.d. and.0 and. mm height for the top Formatted: Font: Bold Formatted: Font: Bold Formatted: Indent: First line: 0 cm

30 and bottom lids, respectively) drilled with a single hole. Two rubber joints placed between the lids and the plastic tube hermetically closed the pressure cell. A first measurement of and a was done with the ceramic disks dry. Next, the WEC P was saturated by injecting a ds m -1 NaCl solution through the base of the pressure-cell. The ceramic disks were considered saturated and equilibrated with the NaCl solution when it exited the top of the pressure cell with the same EC then that used to saturate the WEC P. This process took approximately hours. Next, the ceramic disks of the WEC P were sequentially desaturated at different pressure heads (,,, 0 and 0 kpa) by injecting air through the top of the pressure cell. The extracted water was collected and its EC was measured. The values of and a were recorded at soil saturation and hours following each pressure-head step. This experiment was repeated twice using a ds m -1 NaCl solution. Finally, assuming a negligible a-s (Eq. ), the factor was numerically calculated by minimizing the RMSE between the measured w and the estimated w (Eq. ), for an average sat. This TDR probe was validated in a pressure cell laboratory experiment and under field conditions. The pressure cell consisted of a plastic tube (0 mm i.d., 0 mm height) with a mm i.d. hole drilled at mm height, and closed at the ends with two plastic lids (Fig. 1). The bottom lip had inserted a 0. bar ceramic plate (-mm thick and 0-mm in diameter) (Soil Moisture Inc. UK), which was placed on a mm i.d. hole. These two holes allowed the flow of air and water during the soil wetting and draining processes. Two female-female BNC connectors, in which the WEC P and SWC P were connected, were inserted though the top lip. A thermocouple was also inserted in the pressure cell for soil temperature measurements. The cell was filled up and uniformly packed with sand (0 m grain size) until the head of the TDR probes were half covered. Next, a ds m -1 NaCl solution was slowly injected through the base of the pressure cell until the EC of the outlet solution equalled the inlet one ( hours

31 1 approximately). The total volume of water added using a volume of was approximately four times the total soil porosity. ( hours approximately), and the EC of the outlet solution equalled the inlet one. Once the sand was saturated and equilibrated with the NaCl solution, the column was sequentially drained at pressure heads of 0.,,,, 0 and 0 kpa, by injecting air through the lateral pressure cell hole. The water drained at each pressure head was collected and the corresponding EC measured. Values of and a obtained with the WEC P and SWC P were recorded at soil saturation and hours after imposing each pressure head. According to Eq. (), w was calculated from the measured and a values and the and factors estimated in the previous experiments. The w values were corrected to ºC (Eq. ). This experiment was repeated using a -mm sieved loam soil saturated with three different KCl solutions of, and ds m -1. Finally, the TDR-estimated w values were statistically compared to the measured EC values in the inlet solutions Field testing The field experiment consisted in comparing the w estimated by WEC P to the EC values measured in the soil solution extracted with ceramic tension lysimeters (TL). The experiment was performed on a loam soil located in an apple orchard of the Estación Experimental de Aula Dei (Zaragoza). The soil bulk density was 1. g cm -. Three TL (model SPS 00 - SDEC) were inserted into the soil at the vertices of a 1 cm equilateral triangle, the WEC P was inserted in the center of the triangle, and the SWC P at a cm distance from the WEC P. Both TDR probes were inserted at the same depth that the TL. The heads of the two TDR probes were buried 1 cm under the soil surface. The experimental plot was confined in a 0 cm diameter and 0 cm height plastic tube driven 1 cm into the soil. Successive soil wettingdrainage cycles were repeated with distilled water and KCl-water solutions of different EC Formatted: Indent: First line: 0 cm Formatted: Font: Bold